--- title: README emoji: 🌍 colorFrom: green colorTo: yellow sdk: static pinned: false ---
![olaverse](https://cdn-uploads.huggingface.co/production/uploads/69949cbacd82af728f850c12/gV4cAx9k1eVtIW8FEiONl.png) # Olaverse Lab **Small, task-specific open models — with a core focus on African & Nigerian languages.** We believe the future of open AI isn't only bigger models — it's *specialist* models: compact, efficient, and sharply focused on one job. We build the full stack — language tools, retrieval, generation, and vision — and release it openly, with a mission to make African languages first-class citizens of AI. **30+ open models · Yorùbá · Igbo · Hausa · Nigerian Pidgin — and beyond** [Website](https://www.olaverse.co.uk/) · [GitHub](https://github.com/Olaverse-Labs) · [X/Twitter](https://twitter.com/olaverse_) · [LinkedIn](https://linkedin.com/company/olaverse)
--- ## 🔥 Featured | Model | Task | Why it matters | |---|---|---| | [**diacnet-1.0**](https://huggingface.co/olaverse/diacnet-1.0) | ✍️ Diacritic restoration | Our most-used model — restores diacritics/tone marks across Yorùbá, Igbo, Hausa & more. A ByT5 model doing a job almost nobody else does for these languages | | [**naija-embed-base**](https://huggingface.co/olaverse/naija-embed-base) | 🧭 Nigerian embeddings | Cross-lingual sentence embeddings for Hausa, Yorùbá & Igbo — the foundation for local-language search & RAG | | [**mist-qg-1.5b**](https://huggingface.co/olaverse/mist-qg-1.5b) | ❓ Question generation | Generates questions from any passage in **25+ languages** · [demo Space](https://huggingface.co/spaces/olaverse/Mist_Question_Gen) | | [**lid-neural-5.1**](https://huggingface.co/olaverse/lid-neural-5.1) | 🌍 Language ID | Identifies Nigerian languages — built on our own ModernBERT encoder · [live demo](https://huggingface.co/spaces/olaverse/lid-neural-5) | --- ## 🌍 Nigerian & African Language Stack Over 2,000 of the world's languages are African, yet they remain nearly invisible in AI. We're changing that — starting with Nigeria, one focused model at a time. ### ✍️ DiacNet — diacritic & tone restoration Restoring the marks that make Nigerian-language text readable, searchable, and machine-usable. - [**diacnet-1.0**](https://huggingface.co/olaverse/diacnet-1.0) — flagship multilingual ByT5 restorer (Yorùbá, Igbo, Hausa, + more) - Yorùbá specialists: [diacnet-yor](https://huggingface.co/olaverse/diacnet-yor) (BiLSTM) · [diacnet-yor-x](https://huggingface.co/olaverse/diacnet-yor-x) (AfriBERTa) · [diacnet-yor-viterbi](https://huggingface.co/olaverse/diacnet-yor-viterbi) (statistical) · [diacnet-yor-db](https://huggingface.co/olaverse/diacnet-yor-db) (dot-below) - Igbo: [diacnet-ig](https://huggingface.co/olaverse/diacnet-ig) ### 🧭 Foundation encoder & embeddings A Nigerian-language base model and the retrieval tools built on it. - [mist-encoder-base-ng](https://huggingface.co/olaverse/mist-encoder-base-ng) — ModernBERT masked-LM encoder (Hausa, Yorùbá, Igbo, Nigerian Pidgin) - [naija-embed-base](https://huggingface.co/olaverse/naija-embed-base) — cross-lingual sentence embeddings built on the encoder ### 🌍 Language identification (LID) Two coverage tiers, multiple efficiency options: - **Nigerian-focused (5):** [lid-neural-5.1](https://huggingface.co/olaverse/lid-neural-5.1) · [lid-neural-5](https://huggingface.co/olaverse/lid-neural-5) · [lid-lite-5](https://huggingface.co/olaverse/lid-lite-5) (zero-dependency) - **Multilingual (25):** [lid-neural-25.1](https://huggingface.co/olaverse/lid-neural-25.1) · [lid-neural-25.2](https://huggingface.co/olaverse/lid-neural-25.2) · [lid-lite-25](https://huggingface.co/olaverse/lid-lite-25) (fastText) ### 🔤 Tokenizers - [otk-bpe-50k](https://huggingface.co/olaverse/otk-bpe-50k) — 50k BPE for Nigerian languages (Yorùbá, Igbo, Hausa, Pidgin) - [otk-bpe](https://huggingface.co/olaverse/otk-bpe) — byte-level BPE (Swahili, Kinyarwanda, French, English) --- ## 🔎 Retrieval Stack — Search & RAG A full embed → retrieve → rerank pipeline in compact, deployable pieces: - [naija-embed-base](https://huggingface.co/olaverse/naija-embed-base) — embeddings - [mist-reranker-150m](https://huggingface.co/olaverse/mist-reranker-150m) — ModernBERT cross-encoder for RAG - [mist-reranker-22.7M](https://huggingface.co/olaverse/mist-reranker-22.7M) — tiny reranker for edge/CPU --- ## ⚡ Generators — Small Models, One Job Each - [mist-qg-1.5b](https://huggingface.co/olaverse/mist-qg-1.5b) — multilingual question generation (25+ languages, Qwen2.5-based) - [mist-tg-0.3b](https://huggingface.co/olaverse/mist-tg-0.3b) — 300M ByT5 title generator --- ## 🖼️ Prism — Vision Models Compact image restoration & manipulation: - **Super-resolution:** [prism-upscaler-2x](https://huggingface.co/olaverse/prism-upscaler-2x) · [prism-upscaler-4x](https://huggingface.co/olaverse/prism-upscaler-4x) · [prism-upscaler-max](https://huggingface.co/olaverse/prism-upscaler-max) - [prism-denoiser](https://huggingface.co/olaverse/prism-denoiser) — removes noise, blur & compression artifacts - [prism-steganography](https://huggingface.co/olaverse/prism-steganography) — hide & recover data in images --- ## 🧠 MIST — Our LLM Line Open text-generation models from pocket-sized to frontier-scale, with community GGUF quants available: | Model | Size | Notes | |---|---|---| | [MIST-Mini-8B](https://huggingface.co/olaverse/MIST-Mini-8B) | 8B | Compact flagship (Llama-3.1 merge) | | [MIST-Mini-8B-Thinking](https://huggingface.co/olaverse/MIST-Mini-8B-Thinking) | 8B | Reasoning-tuned (GRPO) | | [MIST-1-70B](https://huggingface.co/olaverse/MIST-1-70B) | 70B | Mid-scale · [demo Space](https://huggingface.co/spaces/olaverse/MIST-1-70B) | | [MIST-1-140B](https://huggingface.co/olaverse/MIST-1-140B) · [4-bit](https://huggingface.co/olaverse/MIST-1-140B-4bit) | 137B | Our largest release (frankenmerge) | --- ## 🗺️ What's Next - 📊 Published benchmarks for every featured model — measured against real baselines, not claimed --- ## 🤝 Get Involved We're a small lab with big ambitions. If you're working on African-language NLP, edtech, retrieval, or efficient specialist models — or want to use our models in production — reach out via [olaverse.co.uk](https://www.olaverse.co.uk/) or open a discussion on any model. *Olaverse Lab — building AI that speaks your language.*